netdev - Re: [PATCH net-next] r8169: perform reset synchronously in __rtl8169

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20180522225155.GB2945@electric-eye.fr.zoreil.com>
Date:   Wed, 23 May 2018 00:51:55 +0200
From:   Francois Romieu <romieu@...zoreil.com>
To:     hkallweit1@...il.com
Cc:     David Miller <davem@...emloft.net>, nic_swsd@...ltek.com,
        netdev@...r.kernel.org
Subject: Re: [PATCH net-next] r8169: perform reset synchronously in
 __rtl8169_resume

David Miller <davem@...emloft.net> :
> From: Heiner Kallweit <hkallweit1@...il.com>
> Date: Mon, 21 May 2018 19:01:19 +0200
> 
> > The driver uses pm_runtime_get_sync() in few places and relies on the
> > device being fully runtime-resumed after this call. So far however
> > the runtime resume callback triggers an asynchronous reset. 
> > Avoid this and perform the reset synchronously.
[...]
> Given what we know about ->ndo_open() and the checks by the callers of
> __rtl8169_resume(), the netif_running() test seems superfluous or
> wrong.
> 
> Either way you need to resolve this somehow.

Actually I do not see why the asynchronous reset would be a problem.

Aforementioned places that use pm_runtime_get_sync() are rtl_{open/close}

1. I understand that rtl_open needs to reach synchronously something like
   a D0 state but it does not need anything else, whence the current no-op
   in the driver runtime suspend/resume handlers (that I should have
   remembered btw).

2. rtl_close() needs the same D0 state to perform the hw counters update
   but it should neither need nor care about rtl_reset_work. rtl_close
   even disables itself the deferred work queue right after the hw counter
   update.

If it's also worth making the code more palatable, more explicit symmetry
between rtl8169_net_suspend and __rtl8169_resume would be welcome.

-- 
Ueimor